Kathpalia, A.; Vlachos, I.; Hlinka, J.; Brunovsky, M.; Bares, M.; Palus, M.

Objective: Finding indicators of early response to antidepressant treatment in EEG signals recorded from patients suffering from major depressive disorder. Methods: Functional brain connectivity networks based on weighted imaginary coherence and weighted imaginary mean phase coherence were computed for 176 patients for 6 different EEG frequency bands. Cross-hemispheric connectivity (CH) and lateral asymmetry (LA) were estimated from these networks based on EEG signals recorded before the beginning of treatment (Vis1) and one week after the start of the treatment (Vis2). Repeated measures ANOVA was used to check for statistically significant changes in connectivity based on these measures at Vis2 w.r.t. Vis1. Post-hoc analysis was performed with multiple pairwise comparison tests to determine which group means were significantly different. Results: It was found that CHVis2 was significantly reduced w.r.t. CHVis1 in the {beta}1 [12.5 - 17.5 Hz] frequency band for the responders to treatment. Also, LAVis2 was significantly increased w.r.t. LAVis1 in the {beta}1 frequency band for the responders. No such significant changes were observed for the non-responders. Brain networks constructed using both weighted imaginary coherence and weighted imaginary mean phase coherence were found to exhibit these results. For the CH connectivity changes, binarized networks and for the LA connectivity changes, weighted networks were found to be more reliable. Conclusions: Responders were found to show a reduction in cross-hemispheric connectivity and an increase in lateral asymmetry, both in the {beta}1 band while no such change was observed for the non-responders. Significance: Decrease in cross-hemispheric connectivity and increase in lateral asymmetry in the {beta}1 band may represent candidate neurophysiological indicators of early treatment response, but they require independent replication before any clinical application can be considered.